Improving the Inhibitory Effect of Phages against Pseudomonas aeruginosa Isolated from a Burn Patient Using a Combination of Phages and Antibiotics

Antibiotic resistance causes around 700,000 deaths a year worldwide. Without immediate action, we are fast approaching a post-antibiotic era in which common infections can result in death. Pseudomonas aeruginosa is the leading cause of nosocomial infection and is also one of the three bacterial pathogens in the WHO list of priority bacteria for developing new antibiotics against. A viable alternative to antibiotics is to use phages, which are bacterial viruses. Yet, the isolation of phages that efficiently kill their target bacteria has proven difficult. Using a combination of phages and antibiotics might increase treatment efficacy and prevent the development of resistance against phages and/or antibiotics, as evidenced by previous studies. Here, in vitro populations of a Pseudomonas aeruginosa strain isolated from a burn patient were treated with a single phage, a mixture of two phages (used simultaneously and sequentially), and the combination of phages and antibiotics (at sub-minimum inhibitory concentration (MIC) and MIC levels). In addition, we tested the stability of these phages at different temperatures, pH values, and in two burn ointments. Our results show that the two-phages-one-antibiotic combination had the highest killing efficiency against the P. aeruginosa strain. The phages tested showed low stability at high temperatures, acidic pH values, and in the two ointments. This work provides additional support for the potential of using combinations of phage–antibiotic cocktails at sub-MIC levels for the treatment of multidrug-resistant P. aeruginosa infections.


Introduction
Antibiotic resistance is currently considered one of the biggest threats to global health [1]. Without immediate action, we are fast approaching a post-antibiotic era in which infections caused by multiple-drug-resistant (MDR) bacteria could kill 10 million people per year by 2050 [2]. Recently, the WHO released a list of 12 antibiotic-resistant bacteria that are the greatest threat to human health. Among them, three are listed as a critical priority for research and the development of new antibiotics, including Pseudomonas aeruginosa [3]. This bacterium is a leading cause of hospital-acquired infections (HAI), specifically in severe burn injuries, which are among the most destructive forms of trauma [4]. Infections of burn wounds represent a major concern, as infections by MDR bacteria such as P. aeruginosa are often life-threatening and hard to cure, due to antibiotic that low doses of antibiotics can be effective in clearing bacteria [32]. Furthermore, these studies have mostly used partially-characterized phages, thus missing critical information about the infection properties of these phages, their host range, and the stability of these phages in care products used to treat wound infections [33][34][35].
Here, we isolated 16 Pseudomonas phages from different environmental sources and selected three based on their performance in the efficiency of plating (EOP) test for further analysis. We characterized the genomic, phylogenetic, and morphological features of the selected three phages, as well as their infection properties and their stability under different temperatures, pH values, and in two standard burn ointments. We finally tested the efficacy of these phages against a P. aeruginosa strain isolated from a burn patient under multiple treatment scenarios-single phages, a mixture of two phages, and a combination of phages and antibiotics at sub-MIC and MIC levels-and determined if phage resistance had developed in treatments. We demonstrated that these phages enhanced the efficacy of the antibiotics used, even at subinhibitory concentrations, against MDR P. aeruginosa. Our work provides support for the potential of using combinations of two phages and one antibiotic at sub-MIC and MIC levels.

P. aeruginosa Strains
Twelve different MDR P. aeruginosa strains that were previously isolated from burn wound patients were selected for phage isolation [36]. We chose P. aeruginosa isolate #14 for our further phage-bacterium interactions study as it was showing a high-level of antibiotic resistance and susceptibility to all three selected phages. This study was conducted following the Institutional Review Board approved studies: IR.UMSHA.REC.1396.923, Hamadan University of Medical Sciences.

Minimum Inhibitory Concentration
We selected four antibiotics of different classes-ciprofloxacin, gentamicin, colistin, and imipenem-commonly used to treat P. aeruginosa for further analysis [6]. The MICs of P. aeruginosa isolate #14 to these antibiotics were determined using the microdilution method [37]. The estimated MICs of the antibiotics for P. aeruginosa isolate #14 were 128 µg·mL −1 for ciprofloxacin, 256 µg·mL −1 for gentamicin, 0.75 µg·mL −1 for colistin, and 2 µg·mL −1 for imipenem. From these four, we chose two antibiotics-ciprofloxacin and gentamicin-that are more available, affordable, and less toxic, especially when compared to colistin [38][39][40].

Isolation of Phages
To increase the chance of isolating phages with high genetic diversity, we sampled three different environmental sources-hospital sewage, a local river, and soil (Rasht: 37 • 16 51" N, 49 • 34 59" E and Hamadan: 34 • 48 0" N, 48 • 31 0" E ). The phages were isolated by mixing 50 mL of the collected water (soils were mixed with phosphate-buffered saline (PBS) at a ratio of 1 to 10 weight to volume (w/v)) with 50 mL of double strength LB and 10 mL of an overnight culture of each target bacterial strain. After 18 h of incubation at 37 • C, 10 mL of the mixture was centrifuged at 4000× g for 15 min, and sterile-filtered through a 0.45-µm membrane filter [41,42]. The filtered phage lysates were checked for phages using a standard plaque assay [43]. The filtrates were diluted in LB at five different dilutions, from 10 5 to 10 9 . Then, 100 µL of diluted phages and 200 µL of target bacteria were mixed with 2.5 mL soft agar, spread on LB agar plates, and incubated overnight at 37 • C [41]. The harvested phages were selected according to their plaque morphology. Phages that showed clear plaques (non-turbid), as a sign to differentiate lytic phages from temperate, were selected and re-isolated by plaque purification from the LB agar plates when multiple phages on the same plate were suspected. Isolated phages were kept in LB at +4 • C for further characterization [41,44].  (3,4,5,6,7,8,9,10, and 11) on the activity of phages was studied by mixing 1 mL of phage with 9 mL of PBS, adjusted at different pH values. The viability of phages was tested after a 24-h incubation at 37 • C, using plaque assays according to Grygorcewicz et al. [45].

Burn Ointments
Phages at a concentration of 10 10 pfu/mL were mixed at a ratio of 1:1 w/v with two standard topical antimicrobial ointments for burn wounds: (1) nitrofurazone that contains nitrofurazone 2%, polyethylene glycol 300, polyethylene glycol 1000, and polyethylene glycol 3000; and (2) silver sulphadiazine that consists of silver sulfadiazine 1%, polysorbate 60, polysorbate 80, glyceryl monostearate, cetyl alcohol, liquid paraffin, propylene glycol, and water [35]. The pH values of these two products are around 6. Nitrofurazone and silver sulphadiazine are the two active compounds of these ointments, with antibacterial and antifungal activity [35]. The phage-containing ointment mixtures were incubated at 37 • C for 2, 4, and 24 h, and the stability of phages was quantified by plaque assay after each time interval. The effective therapeutic titer (ETT) was set at 10 7 pfu/mL, according to Merabishvilli et al. [35].

Single-Step Growth Curve
A culture of P. aeruginosa isolate #14 in mid-exponential phase was infected by phages 6, 32, and 45 at an MOI of 0.1 and incubated with shaking at 37 • C. Then, 500 µL of sample were collected every 10 min, centrifuged, filter-sterilized through a 0.45-µm membrane filter, and kept on ice until titration. For each time point, free phages were counted using a plaque assay. Burst sizes were calculated by dividing the average phage titers of the time points after the burst from the initial average of infecting phage titers [46].

Adsorption Degree
The host bacterium P. aeruginosa isolate #14 was grown to mid-exponential phase, and phages added at a concentration approximately equivalent to a multiplicity of infection (MOI) of five. After five minutes of incubation at 37 • C, the culture was centrifuged, filter-sterilized, and kept on ice until titration. The adsorption degree was calculated by subtracting the number of adsorbed phages after five minutes relative to the phage concentration initially added to the culture.

Phage-Antibiotic Combination
Three phages at a different initial multiplicity of infection (MOI) (0.01, 0.1, 1, and 10) were mixed with P. aeruginosa isolate #14 (~10 8 cfu/mL) and incubated at 37 • C. The enumeration of phages and bacterial cells at time points 60, 120, 240, and 300 min was performed using plate counts (CFU and PFU). For the phage-antibiotic combination conditions, phages and bacterial hosts were mixed at an MOI of 1 with subinhibitory concentrations (1/4 MIC) of gentamicin and ciprofloxacin, and phage and bacterial counts were determined every 60 min for 420 min. We then determined the efficacy of the selected two-phage cocktail at a combined MOI of 1 in combination with the antibiotics at subinhibitory concentrations (1/4 MIC) against P. aeruginosa for an extended period of time, at 4, 8, 12, 24, and 48 h.

DNA Extraction and Sequencing
Phages were enriched using the bacterial culture in the exponential growth phase, and were grown at 37 • C on a shaker incubator for 4-6 h. Phages and bacteria were separated by 30 min of 6000× g centrifugation. The phage supernatants were filtered Viruses 2021, 13, 334 5 of 16 through a 0.45-µm syringe filter to remove the remaining bacteria [11]. The filtrates were centrifuged at 35,000× g for 2 h to collect the phages in the pellets. The phage pellets were resuspended in 1 mL of SM buffer (NaCl/MgSO 4 •7H 2 O/Tris-Cl/H 2 O). The resuspended pellets were treated with DNase I for 2 h at 37 • C to remove non-phage-derived DNA. DNase I was inactivated by heating the suspension at 65 • C for 30 min. Ten µL of 20% SDS and 40 µL of protease K (20 mg/mL) were added to each 500 µL of sample, and incubated for 1 h at 37 • C. Subsequently, 35 µL of 5M NaCl and 28 µL of 10% cetrimonium bromide /0.7 M NaCl were added to the mixture and incubated at 65 • C for 30 min [11]. The aqueous phase was transferred to a new tube, 95% ethanol was added up to two times of the total volume, and samples were incubated overnight at −80 • C. Tubes were then centrifuged at 16,000× g at 4 • C for 1 h [11]. Supernatants were removed, and DNA pellets were resuspended in 100 µL of Tris-EDTA buffer. The extracted DNA was purified using the Genomic DNA Clean & Concentrator, ZYMO Research (Cat No./ID: D4064), according to the manufacturer's protocol. The concentration of the purified phage DNA was measured via Qubit 3 [11,44]. Isolated phage DNA was sequenced using an Illumina MiSeq PE250 at the McGill University and Genome Quebec Innovation Center (http://gqinnovationcenter.com/index.aspx accessed on 1 January 2021).
Bacterial DNA was extracted using DNeasy Powersoil Kit, Qiagen (Cat No./ID: 12888-100) according to the manufacturer's protocol. Isolated bacterial DNA was sequenced using an Illumina MiSeq PE250 at the McGill University and Genome Quebec Innovation Center (http://gqinnovationcenter.com/index.aspx accessed on 1 January 2021).

Genome Assembly and Annotation
Raw reads produced by sequencing were trimmed for the adapters and quality checked using BBDuk (http://jgi.doe.gov/data-and-tools/bb-tools/ accessed on 1 January 2021); default options were used. Trimmed reads were subsequently merged by FLASH [47]. Dedupe (https://docs.dedupe.io/en/latest/ accessed on 1 January 2021) from BBTools was used to remove duplicate reads. SPAdes (https://github.com/ablab/spades accessed on 1 January 2021) was used for the de novo assembly. ORFs were identified for the complete genome using Glimmer (https://ccb.jhu.edu/software/glimmer/ accessed on 1 January 2021) [48]. Identified ORFs were annotated using BlastX (E-value < 1 × 10 −5 ). The analysis was performed within Geneious Prime. The phage genomes have been submitted to NCBI and can be found under the GenBank accession numbers MN563783, MN563784 and MN563785; the bacterium sequence data can be found under the accession number PRJNA595666.

Phylogenetic Analyses
The phylogenetic analyses of amino acid sequences from the major capsid proteins were selected from three isolated phages to construct individual trees. Similar proteins were identified using a blast search with a minimum of 40% sequence identity, over at least 90% of the sequence length. Three conserved phage genes (two structural proteins, the head protein and tail protein, and the phage DNA-packaging protein, the terminase large subunit) were combined to compare similarities among the three isolated phages. Phylogenetic analyses were performed by using the maximum likelihood (ML) method based on the Whelan and Goldman model [49], with 500 bootstrap replicates. The analysis was performed within MEGA. We used progressiveMauve [50] for the DNA level homology comparison between phages.

Transmission Electron Microscopy (TEM)
Filtered phage lysates were centrifuged for 2 h at 25,000× g, and phage pellets were resuspended in PBS. Then, 10 µL of resuspended phage pellets were deposited on carboncoated grids for 10 min and stained with 2% uranyl acetate. The negatively-stained grids were then observed on a Leo 906 E microscope (Carl Zeiss, Germany) at 80 kV.

Phage Isolation and Selection
A total of 16 lytic phages were isolated from different sources, including soil (three phages), hospital sewage (four phages), and a local river (nine phages). We then selected four phages based on their performance in an initial EOP test (Table S1). The selected phages were further genetically characterized. We eventually selected three phages with distinct genetic features and infection properties out of the four initially sequenced phages ( Figure S1). All of the selected phages were able to form plaques on P. aeruginosa isolate #14, although none of these phages were originally isolated against this isolate, as per the readyto-use approach recommended for phage therapy. Phages used in this study were isolated using P. aeruginosa isolates 6, 32, and 45. They were named according to the bacterial viruses guidelines [51]: vB_PaeM_GUMS6, vB_PaeM_GUMS32, and vB_PaeM_GUMS45, but only the last part of these names is used in this paper for simplicity (i.e., 6, 32, and 45).

Phage Stability under Different Temperatures, pH Conditions, and in Burn Ointments
We then assessed the stability of all three phages in two commonly-used burn ointments, nitrofurazone and silver sulphadiazine, for 24 h. Isolated phages were quantified using plaque assays after 2-, 4-, and 24-h incubations with the burn ointments at 37 • C. We observed a significant loss of all phages during the incubation with the ointments ( Figure 1A). We further tested the survival of the selected phages under different temperatures and pH values. The three selected phages were most unstable at 60 and 70 • C, and acidic pH values of 3 and 4 ( Figure 1B).

Transmission Electron Microscopy (TEM)
Filtered phage lysates were centrifuged for 2 h at 25,000× g, and phage pellets were resuspended in PBS. Then, 10 μL of resuspended phage pellets were deposited on carboncoated grids for 10 min and stained with 2% uranyl acetate. The negatively-stained grids were then observed on a Leo 906 E microscope (Carl Zeiss, Germany) at 80 kV.

Phage Isolation and Selection
A total of 16 lytic phages were isolated from different sources, including soil (three phages), hospital sewage (four phages), and a local river (nine phages). We then selected four phages based on their performance in an initial EOP test (Table S1). The selected phages were further genetically characterized. We eventually selected three phages with distinct genetic features and infection properties out of the four initially sequenced phages ( Figure S1). All of the selected phages were able to form plaques on P. aeruginosa isolate #14, although none of these phages were originally isolated against this isolate, as per the ready-to-use approach recommended for phage therapy. Phages used in this study were isolated using P. aeruginosa isolates 6, 32, and 45. They were named according to the bacterial viruses guidelines [51]: vB_PaeM_GUMS6, vB_PaeM_GUMS32, and vB_PaeM_GUMS45, but only the last part of these names is used in this paper for simplicity (i.e., 6, 32, and 45).

Phage Stability under Different Temperatures, pH Conditions, and in Burn Ointments
We then assessed the stability of all three phages in two commonly-used burn ointments, nitrofurazone and silver sulphadiazine, for 24 h. Isolated phages were quantified using plaque assays after 2-, 4-, and 24-h incubations with the burn ointments at 37 °C. We observed a significant loss of all phages during the incubation with the ointments ( Figure 1A). We further tested the survival of the selected phages under different temperatures and pH values. The three selected phages were most unstable at 60 and 70 °C, and acidic pH values of 3 and 4 ( Figure 1B).

Genome Characterization, Annotation, DNA Homology, Phylogeny, and Morphology
De novo assembly of the resulting 339,932 reads of phage 6 resulted in a 66,806 bp genome (G + C content: 55.1%) with a high similarity at the nucleotide level (about 93% pairwise identity over 79% coverage) to the genome of vB_PaeM_C1-14_Ab28, a related phage (Figure 2A and Figure S1). A total of 400,544 reads were produced for phage 32, which resulted in a 280,024 bp genome (G + C content: 36.9%) after assembly using SPAdes (Figure 2A). The genome showed resemblance to Pseudomonas phage phiKZ-a jumbo phage- [28] with 94.4% pairwise identity over 54% coverage ( Figure S1). For phage 45, a genome of 92,218 bp (G + C content: 52.8%) was constructed via the assembly of 346,714 Viruses 2021, 13, 334 7 of 16 reads; this genome is most similar to vB_PaeM_C2-10_Ab02 (97% pairwise identity over 58% coverage) (Figure 2A and Figure S1). In addition, we screened the genome of P. aeruginosa isolate #14 for prophages. Contigs longer than 10 kb, produced by MIRA, were submitted to the PHAST web server (http://phast.wishartlab.com accessed on 1 January 2021) to predict potential prophages. Four prophages were identified, three intact and one questionable.  The three phages were separated into different phylogenetic clades constructed using the major capsid protein. Phage 6 was the most similar to Pseudomonas phages R26 and PB1 for the major capsid protein. Phage 32 was related most to phages phiKZ, and fnug. Phage 45 was most related to phages PA10 and Delftia ( Figure 2B-D). A single phylogenetic tree was also constructed using two structural proteins (major capsid and tail fiber) and one functional protein (terminase large subunit) to compare the similarity between phages 6, 32, and 45. Phages 32 and 45 were more similar compared to phage 6 ( Figure 2E). The isolated phages can be divided into the following phage genus, according to ICTV classification: Pbunavirus, phage 6; Phikzvirus, phage 32; and Pakpunavirus, phage 45.
The transmission electron microscopy analysis revealed a Myoviridae morphology for all three phages.

In Vitro Phage Efficacy
We selected three phages with the highest genetic differences and distinct infection properties (Figures 2, 4A, and Figure S1) out of the four sequenced ones for the in vitro efficacy test. We first tested the activity of each individual phage against P. aeruginosa isolate #14 using four different multiplicities of infection (MOIs) ranging from 0.01 to 10, and including the initial MOI, as phages showed an adsorption degree higher than 95%). Phage 6 showed relatively higher efficiency against P. aeruginosa isolate #14 and was successful in lowering the concentration of the target bacteria more efficiently after 4 h of infection ( Figure 4B-D). The reduction level was significant only when phage 6 was applied at an MOI of 10 after 4 h of infection ( Figure 4B-D).

In Vitro Phage Efficacy
We selected three phages with the highest genetic differences and distinct infection properties (Figures 2, 4A and S1) out of the four sequenced ones for the in vitro efficacy test. We first tested the activity of each individual phage against P. aeruginosa isolate #14 using four different multiplicities of infection (MOIs) ranging from 0.01 to 10, and including the initial MOI, as phages showed an adsorption degree higher than 95%). Phage 6 showed relatively higher efficiency against P. aeruginosa isolate #14 and was successful in lowering the concentration of the target bacteria more efficiently after 4 h of infection ( Figure 4B-D). The reduction level was significant only when phage 6 was applied at an MOI of 10 after 4 h of infection ( Figure 4B-D). To test whether the application of a phage cocktail could improve the inhibitory effect of phages within 5 h of infection and avoid bacterial regrowth, we added the second phage (32 or 45) 4 h after the first phage. We calculated an MOI of 1 for the second phage, based on the bacterial concentration at the time of administration ( Figure 4E), to minimize possible side effects. We also examined whether a combination of phage 6 and two conventional antibiotics (gentamicin and ciprofloxacin, at a 1/4 of MIC) could promote the efficiency of the antibiotic treatment ( Figure 4F). We selected these antibiotics because of their lower toxicity and higher availability. Our results show that treatment outcomes differ based on the type of phage or antibiotic used in combination with Phage 6. We further tested potential synergistic effects between two phages and one antibiotic against P. aeruginosa isolate #14 over an extended 48-hour period. All tested treatment scenarios significantly decreased the concentration of the targeted bacterium by more than one log ( Figure  5A,B). We observed the highest efficacy when phages 6 and 45 were used in combination with gentamicin, lowering P. aeruginosa isolate #14 concentration up to 3 logs after 12 h (see Figure 5A,B). However, this was followed by a rise of phage-resistant phenotypes. Phage concentrations significantly decreased over time in all treatment scenarios ( Figure  5C). We also tested the effect of using lower initial concentrations of P. aeruginosa isolate #14 on the efficacy of the treatment under the same conditions. We treated P. aeruginosa isolate #14 with phage-antibiotic combinations at two different initial concentrations-10 6 and 10 7 CFU·mL −1 . We observed a similar bacterial killing effect with a 2 to 3 log decrease To test whether the application of a phage cocktail could improve the inhibitory effect of phages within 5 h of infection and avoid bacterial regrowth, we added the second phage (32 or 45) 4 h after the first phage. We calculated an MOI of 1 for the second phage, based on the bacterial concentration at the time of administration ( Figure 4E), to minimize possible side effects. We also examined whether a combination of phage 6 and two conventional antibiotics (gentamicin and ciprofloxacin, at a 1/4 of MIC) could promote the efficiency of the antibiotic treatment ( Figure 4F). We selected these antibiotics because of their lower toxicity and higher availability. Our results show that treatment outcomes differ based on the type of phage or antibiotic used in combination with Phage 6. We further tested potential synergistic effects between two phages and one antibiotic against P. aeruginosa isolate #14 over an extended 48-hour period. All tested treatment scenarios significantly decreased the concentration of the targeted bacterium by more than one log ( Figure 5A,B). We observed the highest efficacy when phages 6 and 45 were used in combination with gentamicin, lowering P. aeruginosa isolate #14 concentration up to 3 logs after 12 h (see Figure 5A,B). However, this was followed by a rise of phage-resistant phenotypes. Phage concentrations significantly decreased over time in all treatment scenarios ( Figure 5C). We also tested the effect of using lower initial concentrations of P. aeruginosa isolate #14 on the efficacy of the treatment under the same conditions. We treated P. aeruginosa isolate #14 with phageantibiotic combinations at two different initial concentrations-10 6 and 10 7 CFU·mL −1 .
We observed a similar bacterial killing effect with a 2 to 3 log decrease after 12 h when compared to the higher initial concentration ( Figure 5D). In contrast, we observed a higher efficacy when phages and gentamicin were co-incubated with P. aeruginosa isolate #14 at the early-and mid-exponential phase compared to the late-exponential phase ( Figure 5E).
Viruses 2021, 13, x FOR PEER REVIEW 11 of 17 after 12 h when compared to the higher initial concentration ( Figure 5D). In contrast, we observed a higher efficacy when phages and gentamicin were co-incubated with P. aeruginosa isolate #14 at the early-and mid-exponential phase compared to the late-exponential phase ( Figure 5E). To see whether resistance developed before 24 h, we sampled two additional time points, at 8 and 12 h. We observed that resistance developed after ~12 h of co-incubation of bacteria and phages. We randomly selected a total of 120 bacterial colonies, from time points 12 and 24 h, to test their susceptibility to the combination of phages and gentamicin. For the phage 6 + phage 32 + gentamicin combination, 50% of the selected colonies were still susceptible to one of the two phages after 12 h, in contrast to only 30% of colonies for the phage 6 + phage 45 + gentamicin combination. All the selected colonies for the 24-hour time point were resistant to all three phages ( Figure 5F). We did not detect any changes in the MIC for the selected colonies for either time point. Similar results were observed when we used the MIC concentration. To test whether antibiotic treatment could induce infectious prophages that were predicted in silico, we exposed P. aeruginosa isolate #14 to antibiotics (gentamicin and ciprofloxacin at 1/4 and the MIC). After 1, 2, 4, 24, and 48 h of induction with the different antibiotics at said concentrations, we collected phage supernatants and proceeded with plaque assays on 10 other P. aeruginosa strains (Table S2). We detected more infectious prophages after 2 and 4 h of exposure to antibiotics, respectively. The final MIC did not seem to impact prophage induction, as no clear differences were observed between 1/4 and the MIC (Table S2). To see whether resistance developed before 24 h, we sampled two additional time points, at 8 and 12 h. We observed that resistance developed after~12 h of co-incubation of bacteria and phages. We randomly selected a total of 120 bacterial colonies, from time points 12 and 24 h, to test their susceptibility to the combination of phages and gentamicin. For the phage 6 + phage 32 + gentamicin combination, 50% of the selected colonies were still susceptible to one of the two phages after 12 h, in contrast to only 30% of colonies for the phage 6 + phage 45 + gentamicin combination. All the selected colonies for the 24-h time point were resistant to all three phages ( Figure 5F). We did not detect any changes in the MIC for the selected colonies for either time point. Similar results were observed when we used the MIC concentration. To test whether antibiotic treatment could induce infectious prophages that were predicted in silico, we exposed P. aeruginosa isolate #14 to antibiotics (gentamicin and ciprofloxacin at 1/4 and the MIC). After 1, 2, 4, 24, and 48 h of induction with the different antibiotics at said concentrations, we collected phage supernatants and proceeded with plaque assays on 10 other P. aeruginosa strains (Table S2). We detected more infectious prophages after 2 and 4 h of exposure to antibiotics, respectively. The final MIC did not seem to impact prophage induction, as no clear differences were observed between 1/4 and the MIC (Table S2).

Discussion
The increasing incidence of infections caused by MDR bacteria is a global health concern. P. aeruginosa, a leading cause of nosocomial infections, is one of the three bacterial species in the WHO's list of critical pathogens [2,52]. Phages might provide a better alternative to antibiotics due to their specificity and relative ease of isolation [53]. In phage therapy, multiple phages are usually applied at once as a cocktail to broaden the spectrum of bacterial targets, prevent resistance, and increase efficacy under a collective killing effect [12]. Importantly, phage therapy is not limited to using only phage cocktails, as recent studies have reported that combining phages with antibiotics can also be synergistic [24]. In addition to increasing efficiency, the phage-antibiotic combinations may prevent the development of bacterial resistance and enhance susceptibility to antibiotics [27,54,55]. They can also lead to more acceptance of phage therapy by the medical community. However, despite the huge potential to improve phage therapy success, the synergistic interactions taking place in phage-antibiotic combinations and between phages remain relatively unexplored [23].
Here, we studied the synergetic effects of different phage-phage and phage-antibiotic combinations against a P. aeruginosa strain isolated from a burn wound patient. Consistent with the previous reports [24,56,57], our data indicate synergy between phages, as well as phage-antibiotic combinations against multi-resistant P. aeruginosa, even at sub-MIC and MIC levels. If efficient, combining phages with low doses of antibiotics for infection treatment could help reduce the risk of side effects by antibiotics and be more cost-effective for chronic infections and less antagonistic to the activity of phages [23,31,32]. Yet, caution is warranted as the usage of antibiotics at lower doses without phages has led increased bacterial resistance to antibiotics in vitro [58].
The efficacy of the treatments varied according to the phage-to-bacteria ratio, but generally phages showed higher efficiency at higher MOIs. This is consistent with previous studies showing a positive correlation between high doses of phages (10 7 -10 9 pfu) [13,59], and the efficacy of the treatment [60]. However, the administration of phages at high doses can be a double-edged sword as a high concentration of phages is known to activate the host's immune response, which can lead to the neutralization of the administered phages [61]. We therefore sought to determine whether the bacterial killing effect of phages could be enhanced at lower doses by combining them with other phages and/or antibiotics. We evaluated the killing efficiency of two-phage cocktails, one-phage-antibiotic, and twophage-antibiotic combinations against one strain of MDR P. aeruginosa. Our data show the highest efficacy for simultaneous application of two-phages-one-antibiotic combinations against the P. aeruginosa isolate #14. This is likely due to a high fitness cost encumbered by the bacterial pathogen associated with fighting at multiple fronts, as it has been suggested that coping with multiple stressors can negatively affect bacterial growth [62,63].
As with antibiotics, bacteria can become resistant to phages [64]. One way to limit the development of resistance is to use phage cocktails that include genetically-distinct phages [33,65]. We therefore aimed to determine if a cocktail of genetically-distinct phages, with or without antibiotics, could overcome the development of resistance in P. aeruginosa isolate #14. We observed the survival of a resistant subpopulation after 12 h, which expanded after 24 h. Our findings are in agreement with previous studies that reported a time window of between 6 and 48 h after phage infection for the rise of phage-resistant phenotypes [55,64,66]. Resistance to multiple phages may need several resistance mutations to occur, which increases the associated fitness costs [63]. We tested if resistance to one phage could protect the bacterium against infection by other phages. We found that the resistant colonies isolated from the co-incubation of P. aeruginosa isolate #14 with one of the phages were also resistant to the other phages in the study. This suggests all three phages use a common receptor to infect the bacterial host, and that resistance could have been caused by mutations in the phage receptor on the bacterial surface [67]. Further molecular analyses of resistance phenotypes are warranted to draw a clear picture of the resistant mechanisms used by P. aeruginosa isolate #14 against phages or antibiotics.
The reestablishment of antibiotic sensitivity through phage infection has been previously reported [27]. It has been suggested that phage selection can produce an evolutionary trade-off in multi-resistant bacteria, resulting in increased sensitivity to antibiotics [27]. Evolutionary trade-offs are often observed where organisms evolve one trait that increases fitness, while simultaneously allowing reduced performance in another trait [68]. In our study, we did not observe any changes in the susceptibility of P. aeruginosa isolate #14 to gentamicin after its co-incubation with phages and the antibiotic. Thus, the phages we selected failed to increase the sensitivity to the antibiotics likely due to their inability to impose an evolutionary trade-off in this short amount of time.
We further aimed to determine if the selected phages could be used along with burnwound-care products to enhance antimicrobial pressure during treatment. Many care products are highly acidic antiseptics, which can negatively-impact phage activity [35]. We tested the stability of the phages in two common ointments (nitrofurazone and silver sulphadiazine) over 24 h. Our data show that the incubation of phages with different ointments can result in a significant phage concentration loss after only 2 h. This suggests that phage stability in standard wound care products should be carefully assessed before applying them together. Yet, the application of alginate templates [69] or newly-developed phage hydrogels [70] might provide a solution. One must also consider that most phages can be adsorbed within the first ten minutes of exposure to their host, which might be enough for phages to infect the target bacteria before being inactivated by the acidity of the care products. It should thus be possible to administer phages separately and before applying ointments to give them enough time to infect the target bacteria.
Our results still need an in vivo validation as in vitro resistance might not be relevant in vivo [71]. In a more-complex environment like the human body, bacteria must respond to different environmental stressors, which increases the fitness costs of genetic loss [45,72,73]. Therefore, the evolution of resistance to phages can be more challenging in vivo. In addition, host-associated factors such as the immune response to phages and the pathogenic bacteria targeted can strongly impact phage-bacteria interactions. However, our results highlight that the rational design of phage-antibiotic cocktails to target MDR bacteria can enhance the efficacy of treatment even if low doses of phages or antibiotics are applied. Finding ways to improve the antibacterial activity of phages with lower efficacy can expand our arsenal against multi-resistant bacteria. In addition, as antibiotics are becoming less effective against several bacterial pathogens, combining them with phages is a possible strategy to maintain their clinical effectiveness. To limit the evolution of resistance in bacteria or restore sensitivity to current antibiotics, phage cocktails should carefully be designed to impose evolutionary trade-offs in the targeted pathogen.
Going forward, we are planning to validate our results in vivo as it is highly likely that other aspects of the host, such as the immune response [74] and its structural features [75], will also affect phage-bacteria interactions. Using mouse models of wound burn [76], we will study the efficacy of our phage-antibiotic combinations as well as the underlying development of phage-resistant bacteria.